Robot combat is producing the wrong kind of archive. We have clips, winners, knockdowns and dramatic recoveries. Those are necessary for sport. They are inadequate for engineering.
If humanoid fighting is going to function as an embodied-intelligence benchmark, events need a minimum measurement standard that survives changes in league, robot and control architecture.
Identity and configuration.
Record robot model, hardware revision, team, software build identifier, control mode, operator where applicable, battery configuration, protective equipment, weight and relevant firmware. Without configuration identity, comparison decays quickly.
Action and recovery.
Measure initiated strikes, landed contacts, combinations, counter attempts and time from offensive commitment to recovered stable state. Recovery latency captures the cost of action.
Balance and falls.
Record stumbles, torso-ground contacts, autonomous recoveries, manual resets and time required to regain a valid fighting state. Where telemetry permits, preserve center-of-mass excursions, support margin and joint saturation around each fall.
Intervention.
For autonomous systems, report human interventions explicitly. For teleoperated systems, log control latency and disconnect events. Autonomy should be measured, not inferred from branding.
Energy, thermals and damage.
Report battery state, thermal-limit events, power derating, emergency repair time, replaced components and whether calibration changed after repair. Over a season this becomes reliability data.
Tactical geometry.
Tracking can record ring-control time, distance bands, boundary pressure and trajectory. Statements such as “cut the ring off” can finally be tested against geometry.
Publish the schema before teams optimize it.
Every benchmark changes behavior. The schema should be versioned openly when gaming appears. No single metric should become the sport because combat competence is plural.
Robot fighting is young enough that measurement habits are not fixed and serious enough that the data matter. A minimum open bout record would improve commentary, engineering and accountability while making claims about autonomy much harder to fake.